-
-
Add Audio file
-
Sentiment Analysis for each speakers with happy rating
Inspiration
It all began with a simple yet profound realization: communication is not just about words; it's about understanding emotions, intentions, and the nuances that make each conversation unique. Our motivation stemmed from a desire to dissolve the barriers faced by people with hearing disabilities, who often miss out on these subtleties. We envisioned a tool that could not only decode words but also unveil the hidden emotions behind them, making every conversation richer and more accessible.
What it does
This web application goes beyond simple transcription by analyzing and translating spoken words into text while capturing the emotional essence and psychological nuances conveyed in each conversation. For individuals with hearing disabilities, it acts like a compassionate companion, not just interpreting words but also revealing the feelings and intentions behind them. Upon uploading an audio clip, the system provides a detailed emotional status, psychological insights, and a happiness rating for each speaker, enabling users to understand not just what is said, but how it is emotionally expressed.
How we built it
Building this application was a journey of both heart and mind. We chose Flask for its simplicity and OpenAI's cutting-edge models—Whisper for transcription and GPT-3.5 for sentiment analysis—to ensure we could understand not just words but the pulse of emotions. Every line of code was written with the intent to make emotional understanding more inclusive, storing results securely on AWS S3 and ensuring every interaction with the app was a secure and smooth experience.
Challenges we ran into
Developing our Flask-based audio sentiment analysis application posed several significant challenges. Managing dynamic responses from OpenAI's GPT API required us to devise strategies to standardize the output despite inherent variabilities. We also faced crucial security challenges in handling sensitive keys, such as the OpenAI API key and AWS Access Keys. To mitigate risks associated with hardcoding these into the source code, we implemented a secure approach using environment variables. Additionally, we encountered SSL certificate verification errors with OpenAI's Whisper model and faced hosting limitations due to the large size of the Whisper model, which made cloud deployment challenging due to storage constraints. These challenges highlighted the importance of security best practices and the need for scalable solutions in technology projects.
Accomplishments that we're proud of
We take immense pride in not just overcoming these challenges but in the impact our project has begun to make. We've created a tool that doesn’t just function; it understands and communicates emotions, making it a beacon of inclusivity. Seeing it transform conversations for people with hearing disabilities into rich, emotionally insightful exchanges fills us with immense pride and joy.
What we learned
This project was a profound learning curve about the power of empathy in technology. We learned that when we design with accessibility at the heart, we create something that transcends technological achievement and becomes a lifeline for those it serves. The technical skills were just one part; the true lesson was in the value of emotional intelligence in design.
What's next for Audio Sentiment Analysis
Looking ahead, our dream is to expand this tool’s capabilities to encompass multiple languages and cultural nuances, making it universally accessible. We aim to forge partnerships that can help integrate this tool in educational and professional spaces, fostering environments where everyone, regardless of hearing ability, can fully participate and connect on a deeper emotional level.
Log in or sign up for Devpost to join the conversation.